Synopsis In the early morning hours of 31December 1994, a fire broke out in the conveyor belt system of the AMBASSADOR during the unloading of a cargo of rock phosphate. The fire subsequently spread to the vessel's accommodation, and the combined efforts of the ship's crew and several shore-based fire departments were required to bring the fire under control before it was fully extinguished, some 28hours later. There was no damage to harbour installations, no serious injury and no reported pollution as a result of the fire. The Board determined that, when the conveyors were stopped, a section of one of the conveyor belts ignited, probably because the belt was in contact with an overheated roller. The roller probably overheated due to a bearing failure or to being jammed with refuse which ignited after contacting the overheated bearing. 1.0 Factual Information 1.1 Particulars of the Vessel 1.1.1 Description of the Vessel The AMBASSADOR is a self-unloading bulk carrier with accommodation and engine-room located aft. The basic layout is as shown in the General Arrangement Plan (AppendixA, Figure1). The hull is subdivided longitudinally by four transverse watertight (W/T) bulkheads; a collision bulkhead, one bulkhead at each end of the cargo space and one at the after end of the engine-room. 1.1.2 Description of the Unloading System Unloading is accomplished by opening gates at the bottom of the five hopper-shaped holds and allowing the cargo to fall onto two fore-and-aft conveyor belts which transport the cargo to the transfer and loop belt system at the after end of the cargo space where it is raised to the above-deck unloading boom for discharge over the side. The area in which the belts run under the cargo holds and the space at the after end which houses the loop belt are referred to as the tunnel and the loop belt casing, respectively. The unloading team, under the direction of the chief engineer, consists of the chief officer, senior technician, an operator and three tunnelmen. The operator mans the control room at the forward end of the main deck and monitors the cargo on the discharge boom while the tunnelmen operate the hopper gates to regulate the cargo flow as required by the operator and ensure the safe running of the conveyor system. Communication is by means of radiotelephone (R/T) and a signal light system. All decisions to start or stop the unloading system are taken with the approval of the senior personnel. 1.2 History of Events 1.2.1 Prior to the Fire Cargo unloading started at 1225(3) on 29December 1994, shortly after the vessel berthed at Belledune and was intended to be continuous. However, the rock phosphate cargo was found to be most difficult to retain on the conveyor belts and this, in conjunction with a drop of some 2m between the cargo belts and the transfer belts, gave rise to a lot of dust in the vicinity of the transfer conveyors. There were frequent stoppages because spilled cargo had to be shovelled back on to the belts. Finally, although the tunnelmen were wearing paper face masks, the dust became too dense for work to continue, and, at 0115 on 31December, it was decided to stop the unloading. The gates to No.5 hold, which was being discharged at the time, were closed, and, after inspection of the conveyor system, the crew members left the tunnel area with the intention of returning to resume unloading once the dust had settled. The entire, empty, conveyor system was left running. The operator left his position in the control room and met with the chief engineer on the weather deck to discuss the problem while the tunnelmen proceeded to the control room. A glance through a doorway in the loop belt casing revealed that the dust remained severe, and, because it was believed that the movement of the conveyor system was preventing the dust from settling, the system was stopped at 0205. Fifteen minutes later, at 0220 (after personnel had left the area), a heat sensor in the transfer belt area (below and slightly forward of the loop belt) indicated a fire and, at about the same time, smoke was seen emanating from the open top of the loop belt casing. 1.2.2 Containing the Fire After the general alarm went off, the water sprinkler system in the loop belt casing was started and all mechanical ventilation in the tunnel was stopped. The dense smoke prevented personnel from re-entering the area to determine the location of the seat of the fire and to fight the fire, but water from three of the ship's hoses was directed through doorways in the loop belt casing to supplement the sprinklers in an attempt to drench the transfer belt area below. Since all deck hydrants but one were frozen, two of the three hoses were led from hydrants in the engine-room through the accommodation to the loop belt casing. The master notified the ship's agent of the fire by telephone. The agent and a shore worker, upon becoming aware of the fire, informed the security personnel at a factory in the port, who in turn alerted the fire department and various other community fire departments in the area. The master then contacted the managers, and the owners, who sent two directors to the scene and reportedly set up shore-based response teams to provide assistance in fire-fighting efforts. The fire chief arrived at 0300, followed by a Canadian Coast Guard (CCG) Ship Safety Branch surveyor at about 0530, and the two company directors later in the day. All assisted the master to a varying degree in fighting the fire and coordinating the efforts of the fire-fighters. The fire chief and the two directors remained on the scene throughout. Heat transference through the engine-room bulkheads indicated that the seat of the fire was in the transfer belt area, and its intensification indicated that the application of water through the loop belt casing was unsuccessful. Following discussions between the master, the fire chief and the Ship Safety Branch surveyor, it was decided to attempt direct action by entering the transfer belt area through the hydraulically operated W/T door leading from the engine-room to the lower part of the loop belt casing. At 0625, a party of three persons opened the W/T door. One of them, dressed in standard fire-fighter's clothing and equipped with a spray nozzle hose, entered the loop belt casing. The heat and smoke forced him to withdraw before he could determine the location of the seat of the fire. A couple of areas on the forward bulkhead and loop belt casing in the engine-room later became so heated that paint on the engine-room side began to ignite in patches which was controlled by boundary cooling. The heat and smoke also prevented the W/T door from being closed (the local controls having been rendered inaccessible), but the fire door, which was spring-loaded, closed automatically, and this prevented the fire from directly entering the engine-room. Repeated attempts were made to close the W/T door hydraulically using a remote control on the main deck, but this area had also become too hot for personnel. External fire-fighting was continued and supplemented by pumper trucks directing water through the open top of the loop belt casing. Sea water was also pumped into the empty No.3 hold, running into the tunnel area through the gates and this, in conjunction with the other expended fire-fighting water, flooded the tunnel, eventually immersing the burning longitudinal cargo belts over up to 90percent of their lengths. At 1100, the colour of the smoke began to change from black to grey, and it was believed that the fire had been brought under control. From that time, numerous attempts were made to reach the seat of the fire, but they were always thwarted by heat and smoke. However, at 1320, as no more smoke could be seen, it was believed the fire had been extinguished. At 1330, the ventilation fans were restarted with the intent of removing smoke and fumes so that persons could enter the space and deal with any remaining hot spots. Until that time, the fire had been confined to the tunnel and conveyor belt system. One hour later, at 1430, before personnel had entered the area, dense smoke again emanated from the loop belt casing and all external fire-fighting was resumed. The fire then burned to a greater intensity with heat transference through bulkheads igniting fires in the accommodation, despite pre-cooling with water. It took some 28 hours after the initial outbreak before the fire was fully extinguished; continuous shore assistance consisting of both personnel and material was provided throughout. The CCG vessel SIMON FRASER arrived on the scene shortly before noon on 01January 1995 and stood by. Post-fire examination of the accommodation revealed that some cabin doors were in the open position, which facilitated the spread of fire. 1.3 Injuries to Persons One crew member suffered a minor case of smoke inhalation, and the master had a minor case of frostbite. While attempting to rig a hose on the superstructure, the attending Ship Safety Branch surveyor jumped onto a platform from a height of 0.9m to 1.2m (three to four feet), landed on uneven ice, slipped and injured his foot. 1.4 Damage The vessel's self-unloading system was heavily damaged; both of the transfer belts and their associated controls and fittings were burnt out as was the interior of the loop belt casing. Some 25 to 30m of both the port and starboard fore-and-aft belts was burnt, but the section of the loop belt below the transfer belts was not. The vessel's structure bordering the tunnel in way of No.5 hold was distorted from the heat. A resulting spill of some 200 to 300tonnes of cargo had to be cleared from the conveyor system before the fire site could be inspected. Approximately 25percent of the accommodation was destroyed in the fire with the balance sustaining smoke damage to differing degrees and some water damage. Paint was blistered on some sections of the forward engine-room bulkheads in patches which were common with the loop belt casing. In contrast, a set of oxyacetylene cylinders was discovered toward the after end of the tunnel, unaffected by the fire. There was no damage to shore structures. 1.5 Vessel Certification The AMBASSADOR was certificated and manned in accordance with the requirements for a vessel of her type in the trade in which she was engaged. 1.6 Personnel Certification and Experience The master held an Indian Master's Foreign-Going certificate of competency issued in Bombay in 1986 and also held the appropriate Vanuatu licence. He had been going to sea since 1977, employed by the management company since 1987 and in command since 1990. The chief officer held an Indian Master's Foreign-Going certificate of competency issued in Bombay in 1988 and also held the appropriate Vanuatu licence. He commenced his sea-going career in 1978 and had been serving as chief officer and with the management company since 1989. The chief engineer held an Indian First-Class Motor certificate issued in Bombay in 1982 and also held the appropriate Vanuatu licence. He had been employed as chief engineer since 1982. He had been going to sea since 1974 and employed by the management company since 1986. All of the certificated crew (officers) had been trained in shipboard fire-fighting as required by the IMO Standards of Training, Certification and Watchkeeping. The shore fire chief was head of the emergency team for both the smelter and fertilizer plant. His training had not included instruction in fighting shipboard fires and his knowledge of vessels had been gained when previously employed on the dock. 1.7 Resources Available to Fight the Fire 1.7.1 Vessel The AMBASSADOR was fitted with fire-fighting equipment as follows: Accommodation - 18 dry chemical extinguishers, 16x15.24m (50') lengths of 50mm (2) diameter fire-hose with nozzles, 2 sets breathing apparatus; Tunnel - 2 dry chemical extinguishers, numerous 50mm (2) diameter rubber washdown hoses; Loop Belt Casing - fixed water sprinkler system; Machinery Space - 9 dry chemical extinguishers, 4Halon1211 extinguishers, 6x15.24m (50') lengths of 50mm (2) diameter fire-hose with nozzles. The vessel's fire detection system included two sensors above the transfer belts, both of which activated the same location indicators situated in the wheel-house and in the engine-room. 1.7.2 Shore Three pumper trucks and some 115 persons involved with seven local community fire departments and/or employed at factories in the port area assisted with the fire-fighting activities. There were several fire-hose cabinets containing hoses and hydrants within the port area, but the closest was 150m from the site of the fire, requiring the use of pumper trucks to boost the water pressure. 1.8 Shipboard Fire-fighting Capability 1.8.1 Regulatory Requirements The AMBASSADOR complied with Vanuatu and Canadian regulations as well as with the IMO Safety of Life at Sea (SOLAS) Regulations which exempt ships carrying low fire-risk cargoes from the requirements to carry fire extinguishing systems in cargo spaces. When the vessel was registered in Canada, the ship's owners had approached the Ship Safety Branch of the CCG with respect to upgrading the fire-fighting capability in the self-unloading system. The position of the CCG was that there was no regulatory requirement for a fixed fire-fighting system in the tunnel and associated areas, and that the CCG could not legally approve or endorse the installation of a system that was not required. The CCG stated that it was very positive in its endorsement of the need for such a regulation, but that the intent was to let the marine industry itself steer the process as to the development of draft requirements. It was hoped that the shipowners would thus endorse the resulting regulations which they had developed themselves. Proposals considered by the owners to fit a remote fire-fighting system in the tunnel area were not followed up because there were no regulatory guidelines and because the forthcoming regulations might render the system obsolete. The only action taken was to fit conveyor belting treated with a fire-retardant agent which was unlikely to be affected by new regulations. 1.8.2 Equipment Maintenance Muster lists indicating the emergency duties of each crew member were posted throughout the accommodation, and weekly emergency musters had been held, during which fire-fighting procedures had been explained. However, the crew had not been drilled in fire-fighting which meant that there had been no testing and proving of the equipment. No effective safety equipment maintenance routine was in place, and reliance on equipment standards being maintained was based upon the successful completion of Safety Equipment Surveys, to the requirements of the SOLAS convention and applicable protocols. The most recent such survey had been successfully completed just two months earlier, after the vessel was inspected by a Lloyd's Classification Society surveyor at Caracas, Venezuela. Therefore, the sub-standard condition of the equipment was not expected by the crew, leading to confusion and delays at the time of the fire, with various equipment from throughout the ship being tried and discarded. The ship's fire-hoses in the engine-room were sub-standard. Many were degraded and leaked badly, some were secured to their connectors by metal clips which released under water pressure. One section of hose had male connectors on both ends. A fire plan indicating the type and location of each piece of equipment was posted in the accommodation while another was kept in the ship's office, but they proved to be outdated and did not accurately depict recent changes. The plan in the office burned in the fire while the other, in the accommodation, became covered with soot and indecipherable. Reportedly, there was a plan kept in a weatherproof case near the gangway which was used by the vessel's management to explain the interior layout of the vessel to shore-based fire-fighters. 1.9 Ship-Shore Disparities As they had no training in fighting shipboard fires, the shore fire departments' personnel were not aware of the requirements for a fire plan nor of the non-compatibility of shore and ship fittings, and this led to confusion and delay when attempting to connect the fire departments' hoses to the engine-room hydrants. They were also unaware of the requirements for SOLAS ships to be equipped with international ship-shore connections whereby water from the shore can be directed through the ship's fire mains. In this case, the ship was able to maintain pressure on the fire main. The lack of knowledge by both parties of each other's training in fire-fighting, which applies to totally different circumstances, led to confusion during discussions on the best approach to fight the fire. When the lack of visible smoke seemed to indicate that the initial fire had been extinguished, the ship's personnel believed that water drenching and the closure of the area should be continued until the boundary steel bulkheads and decks had cooled down and had remained cool for a period of time; this would have helped confirm that the fire was fully extinguished. However, shore personnel believed that direct action against remaining hot spots should be carried out immediately; the shore personnel's arguments prevailed and the area was opened and vented, leading to the fire burning with greater intensity and spreading to the accommodation. The Port Authority had not provided the ship with instructions as to the most effective way to alert local authorities in case of emergency; therefore, the master telephoned the ship's agent at his residence to report the fire. Furthermore, the Port Authority's by-laws did not require ships to ensure that a fire plan and other relevant information were readily available to shore-based fire-fighters. 1.10 Possible Origin of the Fire The report prepared by the TSB Engineering Laboratory states that the fire or burn pattern was symmetrical from the engine-room bulkhead forward which strongly suggests that the fire progressed from the aft end of the tunnel forward. Post-casualty inspections showed that the most intense heat had been centred in the area of the transfer belts. By process of elimination, it was determined that none of the electrical equipment or electrical and mechanical machinery had been the cause of the fire. Also, there had been no hot work (burning or welding) recently carried out in the area and there was no reason to suspect arson or smoking. Therefore, frictional heat was considered to be the most probable cause of the fire. The transfer conveyor belting and rollers in the area had been completely destroyed, but testing of similar belts and rollers under laboratory conditions at the TSB Engineering Laboratory determined that temperatures in excess of 400C were required to ignite the rubber-tired steel rollers and 450C to ignite the belting. These temperatures could not be reproduced by the small-scale frictional testing carried out at the laboratory, but the tests did show that it would be difficult for rollers or belting to be ignited by friction alone. It is most probable that an intermediary fuel source with a lower ignition temperature was involved. Indications are that a roller overheated due either to a bearing failure or to being jammed with refuse such as discarded gunny sacking (ignition temperature at the lab of 193C) or nylon securing straps; the overheating causing and sustaining a fire for a period of time. As long as the conveyor system was running, the belting passed rapidly over the heated roller and/or burning refuse but did not ignite. However, when the belting was stopped, the heat source was constant in one area of the belting and sufficient to ignite it. The TSB Engineering Laboratory report reveals that a TSB photographic record of the burn site shows that one of the end rollers on the starboard transfer conveyor was not properly seated in the support bracket, while the adjacent centre roller appeared to be secure. The report explains that this photographic evidence was not noted until after the assembly had been removed and, therefore, the roller was not available for further examination. However, the report continues that typically, a bearing failure results in the collapse of the roller as the bearing support is lost and that as the bracket appears to be still supporting the centre roller, the displacement of the end roller probably required the failure of the end roller's axle. The report concludes that Although the origin and cause of the fire... were not conclusively determined, it is probable that the fire originated in the starboard transfer conveyor and was caused by the heat generated when a bearing failed, seizing and eventually displacing a roller from the idler frame. 1.11 Ventilation of the Tunnel Area The ventilation system consisted of two blower fans in the forward part of the tunnel and two extractor fans at the after end which had proved adequate when unloading less dusty cargoes. There was no dust-removal or air filtration system. Reportedly, the fans were operating, but it could not be determined if they were operating to capacity. In any case, their major function was to provide a continuous flow of fresh air to the area, not to remove or filter out dust. The paper face masks worn by the tunnelmen proved inadequate in providing sufficient filtered air to the wearers. The use of the vessel's compressed air packs to provide the tunnelmen with fresh air had been discussed, but not implemented. The air packs, designed to provide air to the fire-fighters for a limited period of time, are part of the vessel's fire-fighting equipment. When the fire broke out, no decision had been taken as regards putting the air packs to a use other than their designed use. 1.12 Loop Belt Casing Structure Entrances to the tunnel area could be closed off to assist in reducing air flow to the area. However, air was free to flow through the top of the loop belt casing through which the cargo was lifted by the loop belts, and a restricted amount of air could enter through the imperfectly closing hopper gates at the bottom of hopper No.3. Both of these conditions would allow air to provide oxygen to the fire. A W/T door provided access to and from the engine-room to facilitate maintenance/repairs and was intended, when closed, to retain the watertight structural integrity of the ship. The remote-closing position on deck was provided as a precautionary measure should the engine-room become inaccessible following collision or in other emergency situations. A spring-loaded fireproof door had been fitted in conjunction with the W/T door. 1.13 Cargo Rock phosphate is a mined natural product which is not hazardous inasmuch as it is not ignitable, toxic nor hazardously reactive to other materials. In this instance, it was carried in bulk in all of the holds and was in the form of granules, a large proportion of which had broken down into fine dust. This cargo was described as, by far, the dustiest and the most difficult to retain on the conveyor belts of all cargoes which had been unloaded during previous trips. Recent cargoes carried by the AMBASSADOR had been homogeneous but varied, consisting of coal, limestone, gypsum and rock phosphate. Stoppages caused by spilled cargo reduced the rate of discharge to an average of some 240tonnes per hour during the 36hours preceding the outbreak of the fire, whereas continuous discharge could have attained a rate of some 1200tonnes per hour, the maximum rate requested by the shore plant. The vessel's unloading system was capable of delivering 4000tonnes per hour. The conveyor belts can be run at two speeds and are normally run at the faster of the two. The crew regulates the amount of cargo being discharged by adjusting the gates at the bottom of the cargo holds. 1.14 Housekeeping in the Tunnel Area During the investigation, a great deal of discarded material was found in the areas of the tunnel which had not been affected by the fire. This material consisted of gunny sacking and nylon straps used to cover the hopper gates to prevent cargo spillage during the voyage, together with various bits of wood, empty paint cans, etc. 1.15 Vessel Stability In order to assess the effects of the fire-fighting water which accumulated in the cargo conveyor tunnel, the post-casualty stability report(4) analysis focused on establishing the vessel's intact transverse stability characteristics on arrival and also after the fire was extinguished. Comparisons were then made with the regulatory minimum intact stability criteria included in the vessel's Trim and Stability Booklet, as approved by the CCG on 13February 1984. Water used to fight the fire in the accommodation quickly escaped from the various decks via the elevator shaft and, because this water was not retained on the upper decks, it caused no significant rise of the vessel's vertical centre of gravity (VCG). Calculations show that expended water from the fire-fighting plus water used to flood the tunnel and submerge the conveyor system (calculated at some 2244tonnes), together with the filling and pressing up of two ballast tanks with a further 5746tonnes of ballast water, lowered the partially unloaded vessel's VCG and more than compensated for the virtual rise of G attributable to the free-surface effect of the water retained in the tunnel area. The calculations revealed that the vessel's intact transverse stability characteristics on arrival, during the fire-fighting effort and after the fire was extinguished, exceeded the minimum safety criteria as specified in the regulations. Under the still water conditions of the harbour where the ship was not subjected to external wave forces causing rolling or pitching motions, there was no danger from dynamic surging and free-surface effect, and positive initial stability was retained, even though a list of 2.5 developed. There was no tunnel sounding or calibration table to assist in accurately ascertaining the precise weight, or free-surface area, of the water in the tunnel: an estimate made during fire-fighting gave 1000tonnes, but post-casualty calculations based on the water marks throughout the tunnel together with calculations based on the final load condition of the ship, determined that there had been some 2244tonnes of water in the tunnel area. 1.16 Weather While the fire was being fought, there were strong northerly winds with a heavy snow fall and an air temperature of minus13C. The wind chill factor further reduced the apparent temperature by an estimated 10 to 15C. 1.17 Previous Fires There is no past record of a previous similar fire on board the AMBASSADOR. However, TSB records indicate that, since 1975, there have been at least eight fires involving conveyor belt systems on Canadian ships. None of the vessels involved was fitted with a fixed remote fire-fighting system, and each fire required the direct action of personnel. Early detection and prompt extinguishing of localized fires in cargo conveyor systems depend on the presence of personnel in the work area.